sign in sign up
<<
Home , Red supergiant star, Supergiant star, Type II supernova

arXiv:0809.1245v2 [astro-ph] 11 Sep 2008 oudt u understanding. suggest our This theory. sup update as evolution to explode stellar directly current may contradict , progenitor massive results luminou most that expected the evidence among the show are to which given begun t have far which, observations as thus Also supernovae result. supergiants detected Ib/c been type have are free p progenitors IIP, the no supernov type that However supernovae, of importantly predicted. avenue most common conclusions, new most interesting This the many to placed. led be has which can for Curren non-detections, images. 21 pre-explosion pre-explosion another in in with changed identified progenitors This detected was . star the of progenitor remnants first the from backwards working ril umte oRylSociety Royal to submitted Article h emsproa a rtitoue yBae&Zik (1935 Zwicky & Baade by from events introduced such first separated was supernovae term The n r h euto yrgnaceigo otesraeo wh a of The surface deposited. the is layer to enough on thick accreting a after of ignites result that the are and hmi hsrve seHlernt&Neee 00.Cr-olpes Core-collapse 2000). Niemeyer & Hillebrandt (see dwarfs review white this - in of them detonation the are ( supernovae) supernovae, Thermonuclear core-collapse. and thermonuclear atsproa bevdi u aaywr h yh n elrS Kepler G and own Tycho our the respectively. in were 1604 century and a our 1572 twice in or observed once supernovae occurring last only rare, and luminous lwyfdsoe ek n ots oehv ih uv plateau dec four curve luminosity to light their up a days for have brightness few Some constant a months. a over and at light weeks over maximum fades to slowly rise sharp a After ssmlrt h muto nryorSnwl uptoe t 0blinye billion pro 10 the its days over few output a . will over a energy in altered of our significantly amount energy be that must of produce amount To the lifetime. to similar is † h td ftesasta xld ssproa sdt eafo a be to used supernovae as explode that stars the of study The uenvecnb olmnu htte usiealteohrsta other the all outshine they that luminous so be can Supernovae otsbfr aig h nryrqie opwratpclsup typical a power to required energy The fading. before months h oenudrtnigo uenvedvdste notomain two into them divides supernovae of understanding modern The ewrs tr:eouin–sproa:gnrl–sas W stars: – general supernovae: – evolution stars: Keywords: -al [email protected] E-mail: nttt fAtooy h bevtre,Uiest of University Observatories, The Astronomy, of Institute asv tr nterdeath-throes their in stars Massive abig,MdnlyRa,Cmrde B 0HA CB3 Cambridge, Road, Madingley Cambridge, yJh .Eldridge J. John By common .Introduction 1. supergiants oa.Nveocrfeunl norgalaxy our in frequently occur Novae novae. † super loko stp Ia type as know also nveaefrmore far are -novae ed o discuss not do We . l-ae stars: – olf-Rayet htw a need may we that s nyalmto the on limit a only o h hydrogen- the for ,i nunlikely an is s, levariables, blue s er a long has heory l hr r 8 are there tly roa.These ernovae. 97we the when 1987 roadisplay ernova t wr star dwarf ite si galaxy. a in rs proa in upernovae oeiosof rogenitors escstudy, rensic hnthey when ) eio star genitor n remain and lx.The alaxy. research a upernovae T E y and ays Paper X types, ar 2 John J. Eldridge

Figure 1. A massive star’s life cycle. Stars begin their lives on the as compact blue stars. Once the core is formed the surrounding hydrogen envelope expands and the star becomes a red supergiant. Further burning stages occur until an core is formed. The star will explode as a type-II supernova.

account for 72 percent of all supernovae in a volume limited sample (Smartt et al. 2008). They are the final events in the lives of stars more massive than ap- proximately 8 solar (8M⊙). The basic evolution of a star is relatively well understood. A star begins on the main sequence, burning hydrogen to helium. Here it spends the largest fraction of its lifetime. Once a helium core is formed, hydrogen continues to burn in a shell around the core and the stellar radius swells to between 100 and 1000 times that of the Sun. At this stage the star becomes a or, for the most luminous, a red supergiant. Massive stars undergo further burning stages. Helium burns to form carbon and oxygen and then progressively heavier elements burn, preventing stellar collapse until an iron core is formed. Iron fusion is an endothermic reaction. With no further energy source to prevent the core from collapsing a star or a is formed. Figure 1 shows the evolution of a star in schematic form. Creating a releases a tremendous amount of energy in . A fraction of these neutrinos interact with the stellar envelope, providing the energy to heat and eject it and thus give rise to the supernova. The ejected material is rich in heavy elements synthesized during the star’s life and in the supernova. Eventually this material mixes with the interstellar medium and pollutes it. When future generations of stars form they will be more metal rich than the previous generation. Most of the heavy elements in our bodies was formed in the evolution of a massive star and its explosive end. The exact mechanism of how the energy is transfered to the envelope is uncer- tain. Most current simulations of supernova do not produce explosions after the core becomes a neutron star. Different additional mechanisms such as acoustic driving by the proto-neutron star or jets from a magneto-rotational instability as material accretes on to the proto-neutron star have been suggested (see Burrows et al. 2007, Dessart et al. 2008 and references therein). There are two exceptions to the standard picture of iron core collapse. Stars around 7 to 8M⊙ are not massive enough to progress past carbon burning they therefore form oxygen-- cores supported by degeneracy . If the core reaches the Chandrasekhar mass of 1.4M⊙ it begins to collapse. Eventually the central is high enough that are cap- tured by magnesium and neon. This removes the electrons supporting the core and accelerates the collapse to a neutron star and an electron-capture supernova occurs

Article submitted to Royal Society Massive stars in their death-throes 3

Figure 2. The details of the figure are similar to Figure 1. However after the star becomes a red supergiant strong stellar winds (or binary interaction) remove the hydrogen envelope and the star becomes a Wolf-Rayet star. Once an iron core is formed the star explodes as a type-Ib/c supernova.

(Eldridge, Mattila & Smartt 2007; Poelarends et al. 2008). The other exception is the pair-production instability when photons in the core form electron-positron pairs reducing the that was supporting the star and hence lead- ing to its collapse. The resulting evolution is complex and the entire star can be disrupted in an explosion (Heger et al. 2003). However such supernovae are rare, occurring only in the most metal poor (Langer et al. 2007). The appearance of a supernova strongly depends on the structure and compo- sition of the material ejected. This is determined by how much mass is lost from the surface of the star during its evolution. Stars of mass less than 25M⊙ have weak stellar winds and do not lose their hydrogen envelopes before an iron core is formed. Stars more massive than this have strong stellar winds and all hydrogen is lost from the surface before the star explodes. These stars are known as helium stars or Wolf-Rayet stars and Figure 2 shows the structure of such a star. The hydrogen envelope can also be removed if the star is in a binary. The stars in a binary interact if their radius is similar to the radius of the orbit and mass-transfer can occur with one star losing mass and the other gaining mass. Core-collapse supernova progenitors naturally separate into two main groups, those that contain hydrogen when they explode and those that do not. Supernovae are also classified by the same aspect. If hydrogen is detected in the supernova spec- trum then it is classified as a type-II supernova otherwise it is a type-I. These two main types can be further subdivided based on the photometric and spectroscopic behaviour of the supernova (Filippenko 1997). There are two types of type-I core-collapse supernovae: Type Ib with helium lines but no lines in their spectra and Type Ic with neither helium nor silicon lines in their spectra (silicon lines indicate a thermonuclear ). For type-II supernovae, if the luminosity is constant for a few months after the supernova appeared, a plateau in the light curve, then it is a type IIP. These are the most common type of supernovae, making up 59 percent of core-collapse supernovae in a volume limited survey (Smartt et al. 2008). The luminosity is constant because, as the ejecta expand in radius, the visible surface where the observed light is emitted from (the ) moves inwards in mass and therefore remains stationary. There are three other rarer subtypes of type-II supernovae. If the light curve decays linearly then the supernova is a type IIL. The progenitors of these stars are thought to have lost hydrogen from their envelopes so that there is not enough

Article submitted to Royal Society 4 John J. Eldridge to produce the luminosity plateau. If the supernova has narrow hydrogen lines which indicates slow moving ejecta it is a type IIn. This occurs when the supernova ejecta encounter a large amount of material surrounding the progenitor star and are decelerated from a typical ejection velocity of 105kms−1 to 1000kms−1. The last subtype is the hybrid class, type IIb. These are type-II events which metamorphose into Ib supernovae. The progenitors of these supernovae have only the barest trace of hydrogen left at the time of core collapse. The deduction of which stars produce which supernova was a series of educated guesses. It was thought that red supergiants produced type-IIP supernovae and that the other types were produced depending on the amount of mass lost before core collapse. The more mass that is lost from a star, the deeper the layers and the heavier elements that are exposed at the surface. This leads to the stellar type of the progenitor changing from a red supergiant to a Wolf-Rayet star and the supernova type progresses from: IIP → IIL → IIb → Ib → Ic. The amount of stripping depends on the initial mass of the star with the most massive stars losing the largest fraction of their initial mass and exposing the deepest interiors at explosion. The only way to confirm whether this theory is correct is to study the star that exploded in a supernova. This is difficult because the star is destroyed in the supernova. It is possible however that an image may have been taken before the explosion but supernovae are rare. Also only a few galaxies are close enough for individual stars to be resolved in ground-based images. However the Supernovae, 1987A and 1993J, took place in nearby galaxies and it was possible to find the progenitor star in pre-explosion images and determine their nature (see Section 2). The situation dramatically improved with the launch of the Hubble Space Tele- scope (HST). Its resolving power and sensitivity were such that it became possible to resolve individual stars in galaxies out to 60 million light-years rather than out to a few million light-years. This increased the number of galaxies observed in detail and for a few supernovae a year pre-explosion images of the progenitor should exist (rather than a few per century before HST). In 2003 two groups simultaneously found the first progenitor of a type IIP supernova (Van Dyk et al. 2003; Smartt et al. 2004). With the concept tested, many discoveries have followed. The success can be reflected in that, out of the 135 supernovae that occurred from 1999 to 2007 within range of HST, 29 had HST pre-explosion images of the supernova site. The results have confirmed some theories and caused problems for others. In this article we review the study of supernova progenitors, beginning with the first two discovered. This is followed by discussion of the study of progenitors during the HST era, with highlights of the main results of the observations of type-IIP progenitors. We also discuss the progenitors for type-Ib/c supernovae and the growing evidence that type-IIn supernova have progenitors that challenge the preconceptions of stellar theorists.

2. The Supernovae 1987A and 1993J

The first two supernovae with observed progenitors were rare types. In both cases, the progenitor was the result of binary, rather than the better-understood single star, evolution.

Article submitted to Royal Society Massive stars in their death-throes 5

Figure 3. Evolution models of the progenitors of Supernovae 1987A and 1993J. For 1987A (a) the more massive star became a red supergiant, (b) the star expanded and engulfed its companion in its hydrogen envelope, evolution occurred, (c) the companion was absorbed into the envelope of the primary star and the resulting star was a blue supergiant. The core then continued to evolve and the star exploded as a blue supergiant in an unusual type-II supernova. For 1993J (a) the more massive star became a red supergiant first, (b) the envelope grew until surface material was attracted by the companions and mass transfer occurred, (c) the donating star shrank and mass transfer ended. The companion star was then the more massive. The progenitor retained only a small amount of hydrogen and so a type-IIb supernova occurred.

(a) Supernova 1987A Supernova 1987A occurred in the , a satellite galaxy of the Milky way. This made it the closest event to be observed since the invention of the telescope and it was also visible to the naked eye. It was an unusual supernova, spectroscopically similar to a type-IIP supernova but with a peculiar light curve. The progenitor was discovered from photometry and spectroscopy to be a blue supergiant with a small radius, 45 times the radius of the Sun (Walborn et al. 1987), whereas theory predicted it should have been a red supergiant with a radius a few hundred times that of the Sun. We now understand the supernova was unusual because of the small radius of the progenitor but why was the progenitor blue? There are several possible reasons, including low , rapid rotation or binary evolution (Podsiadlowski 1992). The favoured hypothesis today is that the progenitor was the result of two stars merging in a binary. Initially both stars were on the main sequence burning hydrogen to helium. The more massive, a 16M⊙ star, burnt all core hydrogen to helium first and expanded to become a red supergiant. Then between this point and core collapse the size of the primary star became greater than the radius of the orbit and the whole system entered a common-envelope phase of evolution. The secondary star, a 3M⊙ star, was swallowed by the more massive primary to form a single more massive blue supergiant. While this model agreed with the progenitor observation, there was no other evidence supporting the theory of binary evolution. Then, in 1997 a triple-ring system that had surrounded the progenitor became visible after it was ionized by

Article submitted to Royal Society 6 John J. Eldridge the supernova’s ultraviolet flash. Analysis showed that these rings were formed, during the common-envelope phase of evolution, from material that was lost during the merger. This was further evidence for the binary scenario and provided a method to determine that the merger occurred 20,000 years before the supernova (Morris & Podsiadlowski 2007).

(b) Supernova 1993J Supernova 1993J started out as a type-IIb supernova. The progenitor must have lost most, but not all, hydrogen from its envelope before core collapse. The event was nearby in the galaxy M81, 12 million light-years away. Pre-explosion images of the object were consistent with a red supergiant but there was excess blue flux (Aldering et al. 1994). The immediate suggestion put forward was that the progenitor was in a binary system and had a blue companion star. In 2004 the supernova had faded enough for its position to be observed by HST. The blue companion star was found but the red supergiant had disappeared confirming that red had exploded and a binary companion was present (Maund et al. 2004). The binary companion is necessary because, without it, the exploding star would have retained much more hydrogen and have produced a type-IIP supernova. Ini- tially the progenitor was 15M⊙ but lost 10M⊙ of material. After it became a red su- pergiant and its radius became similar to the orbital separation. Unlike the extreme interaction of 1987A, material was transferred to the companion which increased in mass from 14M⊙ to 22M⊙, the remainder being lost from the system (Maund et al. 2004). In the future this star will also explode but it is difficult to predict how the will have affected the resulting future evolution.

3. Supernova 2003gd and other type IIPs After HST was launched in 1990 its archive grew and so did the chance that a supernova would have a pre-explosion image. The first progenitor discovered by HST was the red supergiant progenitor of Supernova 2003gd. The star was remarkable in its normality (see Figure 4, Van Dyk et al. 2003; Smartt et al. 2004). Subsequently a firm progenitor detection was made for Supernova 2005cs (Maund et al. 2005; Li et al. 2006). The two observations confirmed for the first time that the progenitors of the most common type of supernova were the expected red supergiants, as shown in Figure 1. Both were found to have masses of 8M⊙, the predicted minimum mass for a supernova to occur (e.g. Heger et al. 2003; Eldridge & Tout 2004). Other supernovae with available pre-explosion images have less conclusive de- tections and in many cases no progenitors have been detected. However it is possible to place an upper limit on how luminous (and thus massive) the progenitor could have been and yet remain undetected For the IIP progenitors enough observations are available (20 detections and non-detections) to make some statements about the progenitor population, specif- ically the mass range of stars that gives rise to these supernovae. A statistical analysis shows that the minimum mass for a star to explode in a supernova is around 7.5M⊙, while the maximum mass to give rise to a type-IIP supernova is around 16.5M⊙ (Smartt et al. 2008). Therefore we know that type-IIP supernovae

Article submitted to Royal Society Massive stars in their death-throes 7

Figure 4. The progenitor discovery images for supernova 2003gd (Smartt et al. 2004; reprinted with permission from AAAS). Left: the HST post-explosion image showing the location of the supernova in relation to nearby stars. Centre: the HST pre-explosion image showing the location of the progenitor. Right: a ground based Gemini telescope image showing the reason why the HST is required to make these observations. If this image had been used, the star identified as the progenitor would have been a blend of stars A and C in the centre panel. only come from a relatively small range of masses, despite being in 59 percent of core-collapse supernovae. There is a problem that theory suggests that single stars with masses between 16.5 to 25M⊙ should retain their hydrogen envelopes. So what do these stars explode as? One answer is that the hydrogen envelope is not massive enough to produce the plateau in the light curve resulting in a type-IIL or IIb supernova. An alternative is that these stars have cores massive enough for a black hole to form at core collapse and have only a small explosion energy because a large fraction of ejecta material fall back on to the remnant. In principle the resulting supernovae may be dim and difficult to observe. Until progenitors for the other type II supernovae are observed we shall continue to speculate.

4. Type-Ib/c supernovae A massive star can lose its hydrogen envelope in stellar winds or a binary inter- action. The resulting Wolf-Rayet stars are the suspected progenitors of type-Ib/c supernovae (see Figure 2). It is difficult to separate out the type-Ib and Ic progen- itors as both stellar models and observed Wolf-Rayet stars tend to contain helium. In general the more helium-rich Wolf-Rayet stars produce type-Ib supernovae while the highly stripped helium-poor Wolf-Rayet stars produce type-Ic supernovae. There are currently nine type-Ib/c supernovae with pre-explosion images but their progenitors are undetected (Crockett et al. in prep.). If we assume the observed Wolf-Rayet population (van der Hucht 2001) are the progenitors of these supernovae and use the detection limits from the progenitor observations to run a Monty- Carlo simulation we find the probability of nine non-detections is less than 0.05. This suggests that Wolf-Rayet stars are not the only type of progenitor. The most sensitive pre-explosion observation to date is for Supernova 2002ap. The observation rules out a normal Wolf-Rayet star and favours a binary system with a low-mass Wolf-Rayet star with lower mass M ≤ 5M⊙, than a typical Wolf-Rayet star, M ≈ 10M⊙ (Crockett et al. 2007).

Article submitted to Royal Society 8 John J. Eldridge

Low-mass Wolf-Rayet stars lose their hydrogen envelopes in binary interactions. These are stars with an initial mass less than 25M⊙ which cannot become Wolf- Rayet stars by themselves. However such stars have never been observed in our Galaxy. They may be difficult to find because they are less luminous the normal Wolf-Rayet stars or the binary companion required to strip the hydrogen may hide the low-mass Wolf-Rayet star. Nothing can be firmly concluded until a progenitor is observed. Even if the explosion of a normal Wolf-Rayet star is observed, there are still the non-detections to explain. Hence, the low-mass Wolf-Rayet stars in the Galaxy must be found to back up this hypothesis. The only uncertainty is will we be able to recognize them when we observe them?

5. Type-IIn supernovae have LBV progenitors? A supernova is classified as a type IIn when there are narrow hydrogen lines in its spectrum. This indicates that the hydrogen is moving slower than the typical ejecta velocities. These lines occur when the ejecta are slowed through interaction with dense circumstellar material around the progenitor star, such as a dense . The stars with the densest winds are luminous-blue variable (LBV) stars. Rather than having a steady constant wind, these stars are highly variable and can eject more than a in a single mass-loss event, a process that leads to a very dense environment around the star. Traditionally they have been considered transition objects between main-sequence stars and Wolf-Rayet stars with initial mass greater than 60M⊙. The first time LBVs were suggested to be the progenitors of some supernovae was by Kotak & Vink (2006). They were able to model modulations in radio lightcurve of some supernovae by assuming the pre-supernova mass-loss varied as for an LBV star undergoing S-Doradus type variations. Then supernova 2005gl increased interest further in this supernova type and their progenitors. A source consistent with an LBV was in pre-explosion imaging but it is too distant to be sure it was a single star and not a cluster of stars (Gal-Yam et al. 2007). In ad- dition the Supernovae 2006jc, 2006gy and 2005gj had indirect evidence that their progenitors had properties similar to LBV stars (Pastorello et al. 2007; Smith et al. 2007; Trundle et al. 2008). This is an active area of research and the subject is in a state of flux. LBV stars have traditionally been interpreted as objects that have yet to begin or complete core helium burning. Suggesting that they can explode is uncomfortable for most theorists because the stars still have to burn helium and the other products before core-collapse. While these burning stages occur the envelope can be lost and the star becomes a Wolf-Rayet star. A possible approach is to say that most LBVs are still transition objects. We therefore need to ask if there are stars that look like LBV stars but are in fact close to core-collapse. To answer this question we would need to understand what drives the LBV behavior. We do not fully understand this behavior although Stothers & Chin (1996) find that, in the most massive stars with hydrogen envelopes and helium cores, the separates into two quite separate structures, an inner core and a separate unstable outer shell. It is this outer shell that is ejected from the star, leaving the inner core intact. These structures occur in the most massive stars after the end of the main sequence. However a similar structure can be found in a narrow mass range, around 30M⊙, in more evolved stars near to

Article submitted to Royal Society Massive stars in their death-throes 9 core collapse. While typical LBV stars experience the evolution sequence, Main sequence → LBV → Wolf-Rayet → Supernova, these more centrally evolved LBVs experience the evolution of Main sequence → Red supergiant → LBV → Supernova (Eldridge, in prep.). During the red-supergiant phase their cores evolve close to core collapse as their grow. They become LBV stars at the same time as their cores are close to collapse. These stars would differ from the traditional LBVs in their evolutionary status but would appear as LBV stars when observed and be indistinguishable. The luminous and variable red supergiant HV 11423 (Massey et al. 2007) could be a red supergiant evolving towards an LBV phase of evolution . Finally a recent type-IIn Supernova 2008S must be mentioned. While no progen- itor has been observed for this object, a dusty cocoon surrounding the progenitor star was detected. The inferred luminosity indicates that the progenitor was a star with a mass 4 to 7M⊙ rather than a LBV star (Prieto et al. 2008). However there is some debate as to whether the event was a supernova because several expected observational features have not been observed. The true nature of LBV stars and the progenitors of type-IIn supernovae make a complex problem, investigation of which is currently observationally led. More work by theoreticians is required in addition to further detailed observational study.

6. Discussion & Conclusions The direct study of pre-supernova imaging in the archives has led to some quite remarkable and usually understated successes in the determination that the progen- itors of type IIP supernovae are red supergiants. Despite this, it is still a subject in its infancy. The remaining supernova types are still lacking firm constraints. The importance of direct study is highlighted by the current confusion of type-IIn progenitors. By detecting the progenitors of supernovae, we can make fundamental tests of that were not possible before we were able to observe their progen- itors. Each time we see something new it provokes us to reach a new understanding. Our understanding of the lives of stars and the nature of supernovae is evolving dramatically because of this exciting avenue of science. JJE is funded by the IoA Theory Rolling Grant from the STFC. He also thanks Julie Wang, Richard Stancliffe, Elizabeth Stanway, Christopher Tout and the two anonymous referees for the useful and helpful guidance and comments. He also thanks Stephen Smartt and Mark Crockett on continuing collaboration on supernova progenitors.

References Aldering, G., Humphreys, R.M., Richmond, M. 1994 SN 1993J: The optical properties of its progenitor AJ 107 662-672 (DOI 10.1086/116886.) Burrows, A., Livne, E., Dessart, L., Ott, C. D., Murphy, J. 2007 Features of the Acoustic Mechanism of Core-Collapse Supernova Explosions ApJ 655 416-433 Crockett, R. M., Smartt, S. J., Eldridge, J. J., Mattila, S., Young, D. R., Pastorello, A., Maund, J. R., Benn, C. R., Skillen, I. 2007 A deeper search for the progenitor of the Type Ic supernova 2002ap MNRAS 381 835-850 (DOI 10.1111/j.1365-2966.2007.12283.x) Dessart, L., Burrows, A., Livne, E., Ott, C. D. 2008 The Proto-Neutron Star Phase of the Collapsar Model and the Route to Long-Soft Gamma-Ray Bursts and Hypernovae ApJ 673L 43D DOI(10.1086/527519)

Article submitted to Royal Society 10 John J. Eldridge

Eldridge, J.J., Tout, C.A. 2004 The progenitors of core-collapse supernovae MNRAS 353 87-97. DOI(10.1111/j.1365-2966.2004.08041.x) Eldridge, J. J.; Mattila, S.; Smartt, S. J. 2007 Ruling out a massive asymptotic giant-branch star as the progenitor of supernova 2005cs MNRAS 376L 52-56. DOI(10.1111/j.1745-3933.2007.00285.x) Filippenko, A.V., 1997, Optical Spectra of Supernovae &A 35 309-355. DOI(10.1146/annurev.astro.35.1.309) Gal-Yam, A., Leonard, D. C., Fox, D. B., Cenko, S. B., Soderberg, A. M., Moon, D.-S., Sand, D. J., Li, W., Filippenko, A. V., Aldering, G.; Copin, Y. 2007 On the Progenitor of SN 2005gl and the Nature of Type IIn Supernovae ApJ 656 372-381 Heger, A., Fryer, C. L., Woosley, S. E., Langer, N., Hartmann, D. H. 2003 How Massive Single Stars End Their Life ApJ 591 288-300 (DOI 10.1086/375341) Hillebrandt, W., Niemeyer, J.C. 2000 Type IA Supernova Explosion Models ARA&A 38 191-230 Kotak, R., Vink, J. S., 2006 Luminous blue variables as the progenitors of supernovae with quasi-periodic radio modulations A&A 460 5-8. DOI(10.1051/0004-6361:20065800) Langer, N., Norman, C. A., de Koter, A., Vink, J. S., Cantiello, M., Yoon, S.-C. 2007 Pair creation supernovae at low and high redshift A&A 475L 19L DOI(10.1051/0004- 6361:20078482) Li, W., Van Dyk, S.D., Filippenko, A.V., Cuillandre, J.-C., Jha, S., Bloom, J.S., Riess, A.G., Livio, M. 2006 Identification of the Red Supergiant Progenitor of Supernova 2005cs: Do the Progenitors of Type II-P Supernovae Have Low Mass? ApJ 641 1060- 1070 (DOI 10.1086/499916) Massey, P., Levesque, E.M., Olsen, K.A.G., Plez, B., Skiff, B A. 2007 HV 11423: The Coolest Supergiant in the SMC ApJ 660 301 DOI(10.1086/513182) Maund, J.R., Smartt, S.J., Kudritzki, R.P., Podsiadlowski, P., Gilmore, G.F., 2004 The massive binary companion star to the progenitor of supernova 1993J Nature 427 129- 131 Maund, J.R., Smartt, S.J., Danziger, I.J. 2005 The progenitor of SN 2005cs in the MNRAS 364 33-37 (DOI 10.1111/j.1745-3933.2005.00100.x) Morris, T., Podsiadlowski, P. 2007 The Triple-Ring Nebula Around SN 1987A: Fingerprint of a Binary Merger Science 315 1103-1106 Pastorello et al. 2007 A giant outburst two years before the core-collapse of a massive star Nature 447 829P DOI(10.1038/nature05825) Podsiadlowski, P. 1992 The progenitor of SN 1987 A PASP 104 717-729 Poelarends, A.J.T., Herwig, F., Langer, N., Heger, A. 2008 The Supernova Channel of Super-AGB Stars ApJ 675 614P DOI(10.1086/520872) Prieto, J.L., Kistler, M.D., Thompson, T.A., Yuksel, H., Kochanek, C.S.;,Stanek, K.Z., Beacom, J.F., Martini, P., Pasquali, A., Bechtold, J., 2008 Discovery of the Dust- Enshrouded Progenitor of SN 2008S with Spitzer ApJ letters in press. Smartt, S.J., Maund J.R., Hendry, M.A., Tout, C.A., Gilmore, G.F., Mattila, S., Benn, C.R. 2004 Detection of a Red Supergiant Progenitor Star of a Type II-Plateau Super- Science 303 499-503 Smartt, S.J., Eldridge, J.J., Crockett, R. M., Maund J.R. 2008 The death of massive stars - I. Observational constraints on the progenitors of type II-P supernovae in prep. Smith, N., McCray, R. 2007 Shell-shocked Diffusion Model for the Light Curve of SN 2006gy ApJ 671 17-20 Stothers, R.B., Chin, C.-W., 1996 Evolution of Massive Stars into Luminous Blue Variables and Wolf-Rayet Stars for a Range of : Theory versus Observation ApJ 468 842-850

Article submitted to Royal Society Massive stars in their death-throes 11

Trundle, C., Kotak, R., Vink, J. S., Meikle, W. P. S. 2008 SN 2005 gj: Evidence for LBV supernovae progenitors A&A accepted. van der Hucht, K. A. 2001 The VIIth catalogue of galactic Wolf-Rayet stars NewAR 45 135-232 Van Dyk, S. D., Li, W., Filippenko A. V. 2003 On the Progenitor of the Type II-Plateau Supernova 2003gd in M74 PASP 115 1289-1295 Walborn, N.R., Lasker, B.M., Laidler, V.G., Chu, Y.-H. 1987 The composite image of Sanduleak -69 deg 202, candidate precursor to supernova 1987 A in the Large Magellanic Cloud ApJ 321 41-44 (DOI 10.1086/18500.)

Article submitted to Royal Society